The Complete Value-Based Draft Model

In this part of the intermediate series, make a whole value based draft model with Python.

If you have any questions about the code here, feel free to reach out to me on Twitter or on Reddit.

Shameless Plug Section

If you like Fantasy Football and have an interest in learning how to code, check out our Ultimate Guide on Learning Python with Fantasy Football Online Course. Here is a link to purchase for 15% off. The course includes 15 chapters of material, 14 hours of video, hundreds of data sets, lifetime updates, and a Slack channel invite to join the Fantasy Football with Python community.

Completing our Value Based Draft Model

In this part of the intermediate series, we are going to finish up our value based draft model.

In this last part, we talked about how you can use positional value to decide who to draft for two players at different positions with different projected points. Our solution was to calculate a replacement value for each position, find each player's value over replacement, and then finally, sort the table in descending order.

There's one problem with going strictly off this value-based ranking model, though. It's not good enough to draft players at value, we also want to know their ADP so we can attempt to get the biggest bargain on each player. Once we load in our draft model again, you'll see what I mean.

What we are going to do in this post is join ADP data with the value based model, and try to find gaps in ADP rank and value rank. The players with high ADPs and low value ranks are our overvalued players, and the players going later in the draft but should go sooner based off our value model are our sleepers.

Let's import some libraries we'll need first.

And then let's copy/paste the code from last time to get our value model. Again, this analysis is for PPR, but you can find ADP data for standard and half ppr by visiting fantasypros.com. The code will be exactly the same to scrape the data.

Now, let's join this data with ADP data on the PLAYER and POS columns. We can do this using the DataFrame method merge. You could also use join, but the join method requires you to set an index, while the merge method does not.

Let's rank the AVG column to come up with an ADP rank.

Our ADP data is now ready to be merged with our value model.

Awesome, so here, we left joined these two tables together on their common columns. A left join is when you want to keep all of the rows of the left table that don't match in the right table, and drop all the rows that don't match in the right table. Here, our left table is the original df, and our right table is the adp_df. Let's investigate the final df object so we can find some examples of players who have a large gap in ADP and value rank.

How We Can Use ADP To Our Advantage

One example that popped out to me is Robert Woods. Robert Woods has a value rank of 23 but an ADP rank of 46. See, if we were going straight based off our value rank, we might jump right at drafting robert woods when he comes to us at that 20-30 spot in the draft. But if we know his ADP, we know that we can take a calculated risk and wait a couple rounds to draft him later, resulting in an even bargain on the pick.

Let's say we're team 5 in this draft. At round 3, Robert Woods is eligible for us to draft just based off his value ranking. But because we know his ADP is around 46, we can take a risk and wait two more rounds and pick him up in round 5, getting an even bigger bargain.

Not only that, but we can also turn our attention to a player who's value rank is around 20-25 and might not be available in the next 2 rounds or even the next round, given their ADP.

I don't believe in drafting straight off ADP, but I do believe in using ADP in this way to inform your decision on who to draft where. It's not fool proof, someone in your league may reach and pick him up in the next round. But we can still use ADP to inform us of where Robert Woods on average is going to go. In this way, we're using ADP data to our advantage.

Let's calculate sleeper and overvalued ranks based off our df and sort the tables to find those players who want to look out for and potentially get a bargain on, and those players we'd like to avoid (unless they fall to us at their value rank).

These are our top sleepers for 2020. Let's sort the table in ascending order to find our top overvalued players.

And that's it for this post! Our draft model is complete. For reference, I ran this draft model through a FantasyPros mock drafter and got a score of 93, whatever that's worth.

I guarentee if you follow this method of leveraging value scores and ADP together, you will at least be in a good position to make the playoffs after your draft. Championships, I can't guarentee. Championships are won after the draft.

Thanks for reading, you guys are awesome!